( By Richard Sander) For the past fifteen years, the University of California (UC) has operated under statewide policies that formally require race neutrality in university admissions and other operations. In the years leading up to 1997–98— when the race-neutral policies went into effect—many parts of the university had been unusually aggressive in using race to increase the campus presence of underrepresented minorities (URMs, comprising African Americans, Latinos, and American Indians). Because of this sharp shift in policy, because of the sheer size of the university, and because its eight principal campuses vary significantly in their academic eliteness, UC presents a rich opportunity to study the effects of a preferences ban.

One of the most striking effects of formal race neutrality across the UC system was a jump in the interest of administrators and many faculty members in the use of socioeconomic status (SES) metrics as an alternative to race in pursuing campus diversity. This chapter will give a brief but substantive overview of how the SES alternative played out. First, I will examine these efforts at the undergraduate level from 1997–2001; second, I will discuss a particularly ambitious program at the University of California, Los Angeles (UCLA) Law School (which I helped develop); and third, I will consider some key changes in undergraduate admissions since 2001.

Undergraduates (1997–2001)

Even before the arrival of the ban on racial preferences, the undergraduate campuses at the University of California had unusually high levels of socioeconomic diversity. At elite private colleges in the 1990s, generally no more than 10 percent of students came from households in the bottom half of the socioeconomic distribution, and even at elite public universities, like the University of Virginia or the University of Michigan, representation of the “bottom half” appears to have been less than 15 percent. At Berkeley and UCLA—the two most elite UC campuses—the comparable number in the mid-1990s seems to have been in the 25 percent to 30 percent range. There were a few reasons for this. The UC system had very low tuition and did an unusually good job of providing grants and loans to students with financial need. California had (and has) a large Asian immigrant population, many of whose teenagers had both low SES and relatively strong academic performance. And the presence in California of a very large and often low-income Hispanic population—many of whom were also recent immigrants—meant that racial preferences in the UC system tended to more effectively reach low-SES households than do racial preferences in most of the United States, where they often disproportionately benefit relatively affluent blacks.

Formal race neutrality nonetheless pushed UC much further in this direction. Indeed, at the same July 1995 meeting at which the UC Regents launched the race-neutral mandate, it established an Outreach Task Force to identify methods through which the university would remain “accessible to students of diverse backgrounds.” After the task force reported back in 1997, the university set in motion a series of outreach strategies aimed at helping high school students “to overcome educational disadvantages” and attract to UC a “student body broadly representative of the state.” Various UC campuses established partnerships with struggling high schools and feeder junior high schools; UC expanded pilot programs aimed at preparing students for UC entry in general and STEM careers in particular; and the university created new information strategies aimed at making students more aware of the curricular steps they would need to take to qualify for admission.

Nearly all UC campuses also modified their admissions policies to increase student diversity. An analysis by economists Kate Antonovics and Ben Backes compared freshman admissions at the eight undergraduate UC campuses in the three years before and the nine years after the ban on racial preferences went into effect in 1998. They found that every UC campus significantly changed its admissions process in the race-neutral era, and that every observable change had the effect of cushioning the drop in minority admissions. Some of these changes were academic: most of the eight campuses decreased the weight given in admissions to standardized test scores (for example, the SAT I) and increased the weight given to high school GPA. Others were socioeconomic: six of the eight UC campuses increased the admissions weight given to parental education (those whose parents had less formal education received a boost) and five of the campuses increased the weight given to students with lowerincome parents.

These initial efforts came without any change to the university’s basic underlying admissions rule: California students who finished in the top eighth of California seniors (as measured by a combination of test scores and high school grades) were guaranteed admission to at least one UC undergraduate program. Racial preferences had been used to create “special admissions” for several hundred URM students each year, but that pipeline essentially disappeared with the arrival of race neutrality in 1997–98.

Remarkably, however, the university managed this multi-faceted shift—eliminating special minority admissions and very large racial preferences while de-emphasizing test scores and substantially expanding socioeconomic preferences—without either a large decrease in overall minority enrollments or a decline in the academic credentials of students. Systemwide, black freshman enrollment fell from 917 in 1997 to 832 in 2000 (a drop of just under 10 percent) and Latino freshman enrollment rose from 3,131 to 3,479 (a rise of just over 10 percent). Meanwhile, median SAT scores held steady and high school GPA levels went up significantly at every campus. There seem to be five explanations for this successful balancing act:

1. Applications from students of all races increased sharply after the adoption of formal race neutrality. The largest year-to-year increase in freshman applications in UC history (up to that point) occurred in 1998, the first year of race neutrality, and overall the number of unique UC applicants was nearly 20 percent higher in 1998–2000 than in 1995–97. This meant that UC schools could be less academically selective while still maintaining very high-credentialed freshman classes.

2. Enrollment rates (that is, the percentage of admitted students who accepted) jumped sharply for URMs with the arrival of race neutrality. The jump was as great as 15 percent for campuses, like Berkeley and UCLA, which had used particularly large racial preferences until 1998. The most plausible interpretation is that students of all races—and especially URMs—were attracted to schools that had abandoned aggressive affirmative action.

3. The shift in admissions criteria, according to the Antonovics and Backes calculations, blunted the racial impact of race neutrality by about one-third. It is important to note, for purposes of this volume, that SES preferences alone—especially the simple preferences used by the UC campuses during this period that looked at factors like parental income and education, not neighborhood poverty concentrations and wealth—would, by themselves, have only done a little to offset the loss of racial minorities from the shift in racial preferences.

4. The UC structure—with eight campuses of significantly varying eliteness—also helped blunt the effect of race neutrality. During the era of large racial preferences, Berkeley and UCLA had lured a disproportionate number of black and Hispanic admits to those campuses; after 1998, many of these students “cascaded” to less elite campuses. This did not produce a concentration of minorities at the least elite campuses, but rather evened-out the distribution of URMs across the eight campuses.

5. Though it is difficult to prove, there is reason to think that many UC campuses continued to surreptitiously rely on race or racial markers in making decisions. The use of race undoubtedly fell dramatically after 1997, but in regression analyses predicting admissions, race continued to be a significant factor.

These various changes at UC were accompanied by a surge in student graduation rates, especially for URM students. All the UC campuses awarded an average of 802 bachelor degrees to African Americans from 1997 through 2003; these were in general the last cohorts admitted with large racial preferences. From 2004 through 2009, in contrast, the UC campuses awarded an average of 926 bachelor degrees to African Americans, despite the initial post-209 drop in black enrollment. Moreover, many more of these students were completing their degrees in four years. Latinos experienced even larger gains. According to the most comprehensive study of these effects, some of this sharp improvement (especially in the sciences) occurred because URM students were better academically matched at the campuses they attended. But part was due to an apparently greater focus by the colleges on assisting students, by providing better counseling, offering more sections of needed courses, and so on. The implication of this latter finding is that administrators pay more attention to minority success when it is harder to achieve the desired racial balance simply through the use of racial admissions preferences.

Over time, the outreach efforts launched at UC campuses also had an important effect in expanding both racial and socioeconomic diversity. The number of California high school applicants to UC from lowor-moderate-income families more than doubled between 1995–97 and 2003–05, even though this was a time when the UC budget was in turmoil and tuitions were rising sharply.

These were the major short-term effects of race neutrality and the university’s greater focus on SES disadvantage. After 2001, the university’s strategy shifted in significant ways—a story I will return to in Part III.

UCLA Law School

Meanwhile, the various graduate schools and departments in the UC system were making their own accommodations to official race-neutrality. UCLA Law School (UCLAW) was unique in developing an experimental program in class-based affirmative action that was comprehensive, based on social science research, and rigorously evaluated. The program was inaugurated in 1997 (for a variety of administrative reasons, only undergraduate admissions were permitted to wait until 1998 to be subject to race-neutral rules) and arose after long faculty debate over how to modify its admissions policy (which were, like those of nearly all law schools, heavily race-conscious) to comply with the new regime. In the new program, UCLAW sought to create a purely “objective” method of assigning weight to student applicants based on their level of SES disadvantage.

The school asked students eight optional questions about their background: the educational level of each parent, their parents’ income and net worth, the applicant’s home address during elementary and high school, and the location of the applicant’s chief elementary and high school. The first four questions pertained to the applicant’s family SES; the last four aimed to measure the SES of the applicant’s neighborhood and school environment. The school questions proved impractical to use; although one could get good data on the socioeconomic composition of California public schools, about half of the law school’s applicants came from outside California or had attended private high schools. The question about home location during elementary school was not used because only 72 percent of the applicants answered it, compared with 88 percent reporting their home address during high school. These addresses were fed through a program that matched them to unique census tracts, the small geographic units for which the census reports detailed neighborhood data. Drawing on the census, UCLAW assigned to each applicant three measures of neighborhood SES: the proportion of neighborhood families headed by single parents, the proportion of neighborhood families receiving welfare, and the proportion of neighborhood adults who had not graduated from high school.

There was nothing especially magical about the particular neighborhood factors used in UCLAW’s system, but there was a clear general rationale for the system. A good deal of social science research suggested that “neighborhood” as well as “family” disadvantage affected life chances; family conditions counted for somewhat more, but not a lot. The three neighborhood factors mentioned, along with the four family measures, each had been used by various social scientists as ways of showing the effect of disadvantage on the later outcomes of young people growing up. Giving some weight to each of these factors seemed both reasonable and objectively justifiable. The law school faculty was also very cognizant of the racial dividends of using broad rather than narrow measures of SES. The correlation between family income and being African-American, in a typical pool of college applicants, is under 0.2; but blacks with middleclass occupations and incomes tend to have significantly fewer assets than do otherwise similar whites, and because of housing segregation, they tend to live in significantly less affluent neighborhoods and have more poor neighbors. The correlation between race and a broader measure of socioeconomic status is thus substantially higher (closer to 0.4), and it makes sense on both opportunity and diversity grounds to measure SES more comprehensively.20

UCLAW thus ended up with seven distinct SES measures used in “scoring” applicant backgrounds. For each measure, the school calculated the mean and standard deviation of the applicant values. Applicants were eligible for a fifteen-point boost if they placed in the most disadvantaged sixth of applicants on any measure—that is, about one standard deviation below the mean on measures like income, or one standard deviation above the mean on measures of neighborhood distress. The boost on any measure could go as high as forty-five points for an applicant in the most disadvantaged 0.3 percent (that is, three standard deviations from the mean). The maximum number of “SES” points any student could receive was just under 200.

UCLAW had used for some years an academic index scaled from 0 to 1000, which gave roughly equal weight to LSAT scores and a student’s “national grade” (college grades adjusted for both the difficulty of an applicant’s college and the degree of grade inflation at the college). An academic index of around 800 was sufficient to gain admission to the school; the median number of SES points received by an applicant for whom they made a difference was about 40; such an applicant had an LSAT score only a couple of points below the class median.

UCLAW’s socioeconomic preferences were, in size and scale, quite different from its old, racial regime. Racial preferences had generally been used for 20 percent of the class or less, but the typical beneficiary of racial preferences received a boost equivalent to 7 or 8 LSAT points (the difference between a ninetieth percentile score and a seventy-fifth percentile score). SES preferences, in contrast, were used for slightly over half of the 1997 first-year class, but, as noted, tended to be much smaller. Race preferences had been narrow but deep; SES preferences were broader and generally shallower.

But although the SES preferences tended to be modest, they were targeted at students who were substantively quite different from the typical elite law school student. Nationally, elite law schools (then and now) draw only about one-tenth of their students from the bottom half of the national SES distribution, but from 50 percent to over 80 percent of UCLAW’s SES preference beneficiaries came from the bottom half of the national SES distribution (depending on which metric is used). As a result, the impact on the school’s demographics was profound. The median family income of first-year students fell from (in 2013 dollars) roughly $140,000 to about $75,000; the proportion of students who were the first in their families to attend college roughly tripled.

A different way of putting this—and a useful way for schools to measure the degree to which they achieve socioeconomic diversity—is to use the index of dissimilarity to compare the distribution of student SES with the national distribution. The index of dissimilarity is often used to measure how different two distributions are, and was originally developed to measure levels of housing segregation. If we were comparing the income distribution of two groups, a measure of 1.0 would indicate there was no overlap in the income of the groups, while a measure of 0 would indicate that two groups had an identical distribution. The income dissimilarity between UCLAW students and the national population fell with the introduction of SES preferences, from about 0.35 to about 0.12; the index of educational dissimilarity fell from about 0.45 to about 0.15.

The racial effects of UCLAW’s socioeconomic preferences were sobering but not surprising. Black enrollment at the school fell by nearly 50 percent from 1996 to 1997; Hispanic enrollment fell by about 15 percent. Asian enrollment slightly increased, so the first-year class was over one-third nonwhite. As with UC generally, the fall was cushioned by a significant increase in yield rates among URMs—suggesting, again, that many minority applicants preferred to attend a school that did not award them a racial preference.23

These declines were not surprising because of the nature of the dilemma facing any highly ranked graduate program. To an overwhelming degree, UCLAW’s competitors for students were other “top 25” law schools around the country, all but two of which (UC Berkeley and University of Texas) were free to continue using racial preferences even as they were eliminated at UCLAW. This meant that any black student admitted by UCLAW strictly on academic credentials would also have offers at far more elite law schools (such as Harvard, Yale, and Stanford), and these offers would often come with generous race-based scholarships. This had been true throughout the era of racial preferences at UCLAW, and continued to be true in 1997—but now UCLAW was admitting many fewer blacks with low academic credentials. Thus, even though the school’s black yield rate went up, it still lost most of its strongest admits to schools using racial preferences. These same dilemmas affected Latino admissions, though the negative effects were smaller with Latinos because the size of racial preferences had long been smaller and a larger proportion of Latino applicants came from low-SES backgrounds. Note here the important contrast with UC’s undergraduate admissions: the key competitors of these colleges were the other seven UC campuses, who were under the same ban on racial preferences. Given the uneven playing field UCLAW faced, with competitors using large preferences, the decline in minority enrollments would have been greater without its class-based affirmative action.

The socioeconomic effects of the 1997 experiment were, then, outstanding, and the racial dividends were substantial. The experiment performed well under other criteria as well. Academically, the students receiving SES preferences performed at the level predicted by their entering credentials, and with “smaller, broader” preferences, the number of students in academic difficulty fell sharply. UCLAW had its historically highest California bar passage—90 percent—from the “class of ’97” compared to rates of 82–87 percent during the preceding years, when the school used racial preferences. The law school also found the financial aid challenge quite manageable, partly due to other favorable developments. Prior to 1994, much of UCLAW’s financial aid had been distributed by lottery to anyone who applied. But in that year, the UC began to allow its professional schools to charge a higher tuition than applied to undergraduates. This “differential fee” was initially quite modest ($6,000 in the late 1990s), but from the outset, schools were required to devote a full third of the fee to financial aid. UCLAW consequently overhauled its financial aid policies and fully implemented a predominantly need-based system by 1995. The revenues from the fee were enough to provide significant aid; and this helped propel a more general increase in yield rates when SES preferences arrived in 1997.

The Second Phase

A very striking characteristic of UC’s academic environment in the Prop 209 era has been an extraordinary administrative distaste for race neutrality, and a tendency to see the consequences of race neutrality as unabashedly negative. Even as UC administrators were overseeing significant increases in socioeconomic diversity, and dramatic improvements in the academic outcomes of students during the 1998–2000 period, they did not publicly acknowledging these achievements and, instead, generally focused official commentary on the declines of URM enrollment at the system’s most elite campuses. The unwillingness of campus administrators to concede any positive effect from race neutrality greatly chilled thoughtful discourse about the extraordinary experiments underway.

Under a steady drumbeat of administrative pressure to increase racial diversity among freshman admits, UC campuses implemented several initiatives starting in 2001. For example, in that year, on the recommendation of UC’s senior administrators, the UC Regents adopted Eligibility in the Local Context (ELC), a new path to UC eligibility quite similar to the Top 10 Percent plan adopted in Texas during the Hopwood years (see Marta Tienda’s chapter in this volume). Under ELC, students at any California public school who ranked in the top 4 percent of their class were UC eligible; in 2011, this was expanded to include the top 9 percent. Of course, at an average high school, the top 4 percent of the class would already be UC eligible under the existing policy of admitting the top eighth of California seniors; so the plan as a practical matter was a race-neutral way of capturing more Hispanic and black students attending inner-city schools whose graduates were underrepresented at UC. It clearly had this effect, and those expanding the pool were also more likely to come from low-to-moderate SES backgrounds.

Meanwhile, some undergraduate campuses were moving beyond simple metrics such as parental income and education in assessing socioeconomic disadvantage. UCLA, for example, was assigning to each applicant a “Life Challenges” score based on an overall assessment of a student’s file, including both objective and subjective factors. We do not know if this measure did a better job of capturing true disadvantage, but we do know that it was more highly correlated with race than the old measures had been.

Finally, many campuses and programs—especially those at the most elite campuses—adopted new admissions practices that were hard to distinguish from racial preferences. UC Berkeley Law School (also known as Boalt Law School) announced in 1998 a new policy of evaluating applicant disadvantage, but it was open secret within and beyond the school that faculty committees were effectively conferring racial preferences—often quite large ones. At UCLAW, many faculty members were dissatisfied with the decline in black and Hispanic enrollment from the 1997– SES experiment. In 1998, the school introduced a more discretionary approach to evaluating SES, but this produced an even larger decline in black enrollment. Then, in 2001, UCLAW created a special admissions track for students interested in Critical Race Studies. The very nature of the program implied a subterfuge, and available data from the initial years of this program suggests dramatic discrimination against white applicants.

UC Berkeley’s undergraduate program adopted a “holistic” admissions process in 2002, which hired special readers to assess all aspects of applicants in a single score that, in the eyes of some involved in the process, encouraged racial preferences (and was documented to have a mildly pro-black effect). In 2006, in the wake of protests at UCLA over the absence of more blacks in the freshman class, that campus developed its own version of holistic admissions, which included a “supplemental review” process that had a dramatic disparate effect in admitting more African-Americans. In 2011, in response to incidents at a couple of campuses indicative of racial tensions, UC-wide policy committees directed all the undergraduate campuses to adopt something like UCLA’s version of holistic admissions.

Concurrent with all these shifts, however, were signs that UC’s late 1990s strategy of improving high school outreach and partnering with troubled high schools was paying dividends. The rate of high school completion for young adults in California rose sharply from 2000 to 2010, as did the rate at which high school graduates completed the core courses required for UC admission. Both trends had a powerful effect—far greater than the use of race-conscious subterfuges—in fueling a near-doubling of Hispanic freshman enrollment on UC campuses over these years.

The UC experience thus illustrates a few different themes. One is that a formal ban on racial preferences prompts experimentation in the use of socioeconomic preferences. This experimentation produced immediate, substantive changes at UC campuses, of which probably the most dramatic and best-documented was the UCLAW experiment. A second theme is that these policies did increase and enhance socioeconomic diversity, produced significant racial dividends, and were consistently accompanied by improved student outcomes. Looking at the UC undergraduate campuses as a whole, the racial effects seemed particularly appealing—URM students cascaded to campuses where they were better matched, while outreach programs increased URM applications and other curricular efforts contributed to the boost in graduation rates. As noted earlier, the UC system was producing many more black and Hispanic graduates, in more challenging majors, during the early years of race neutrality than it ever had before.

A third theme is that race-neutral policies are not self-executing. Implementation and enforcement mechanisms matter. More transparency in admissions and about student outcomes is important for creating accountability. In many university environments, discussions of diversity begin and end with a mere nose count of black and Hispanic numbers in the freshman class; it should not be surprising that university administrators respond to the simple incentives of a one-dimensional standard of success. Making the conversation about university diversity embrace class as well as race, outcomes as well as entering numbers, is vital in creating a culture where innovative admissions policies can flourish.